Who Will Win the Nobel Prize in Economics? Here’S a Forecast Based on Citation Indicators

Home , Martin Feldstein, Trygve Haavelmo

EUGENE GARFIELD INSTITUTE FOF3 SCIENTIFIC IN FORM AT ION8 3501 MAFIKETST Pnil_AOELPHt A PA 19104

Who Will Win the Nobel Prize in Economics? Here’s a Forecast Based on Citation Indicators

Number 11 March 12, 1990

Speculations on who will win the Nobel Prize have appeared in the popular press. 7he Scierrrist~’s success in forecasting the 1989 prize in physiology or medicine is discussed, as is the correlation between citations, receipt of Nobel Prizes and other prestigious awards, and subjective expert judg- ment. Baaed on primary author data in the 1966-1986 .%&d Sciences Citationbukrm, the 50 most-cited economists “of Nobel class” are identified and compared with peer judgments.

Forecasting the Nobel Prizes be fairly accurate indicators of Nobel stature. The Lasker Award and the Gairdner Recently, the newspaper The Scientist@ Award are two of these so-called Nobel pre- published a list naming 20 researchers who, dictor prizes. Further, prizewinning capacity it predicted, were $‘of Nobel class” and and big citation tallies have been found to most likely to win the prize in physiology be highly correlated.4 or medicine. I About two weeks later, the So, 7he Scientist used as its selection cri- 1989 award was amounted and, as it hap- teria exceedingly high citation counts (pick- pened, two scientists on the list were select- ing names from a list of the 500 most-cited ed by the Nobel committee: J. Michael Bish- scientists of 1973-1984), receipt of Nobel op and Harold E. Varmus, both of the Uni- precursor prizes, and preferably both, to for- versity of California, Sart Francisco. mulate its list of 20 scientists who are “of What was the secret of% Scientist’s suc- Nobel class” and likely to become laureates cess in forecasting the prize-” inside infor- in future years. mation, ” expert informants, or just dumb The only surprise was how quickly this luck? It was none of these. Long ago it was prediction came true. demonstrated that scientists who win Nobel Prizes are typically highly cited—about 30 Ficking Winners in Economics times more often than the average scientist.z For example, in 1968 ISI” compiled With the awarding last October of the a list of the 50 most-cited scientists in the 1989 Alfred Nobel Memorial Prize in eco- 1967 Science Citarion [email protected] list in- nomic sciences to University of Oslo, Nor- cluded six scientists who had already won way, professor Trygve Haavehno, the Nobel the Nobel Prize and six more who were committee paid tribute to a pioneer of the awarded the prize in later years. use of statistical probability in economic In addition to citation data, it is well analysis. In 1941 Haavehno wrote a doctoral known that many Nobel Prize winners in dissertation at Harvard University, Cam- physiology or medicine have previously bridge, Massachusetts, entitled “The Prob- garnered other awards that have proven to ability Approach in Econometrics. ” The

83 Nobel committee’s citation singled out this next Nobel Prize. Romer gave Passell the work, which, it stated, “had a swift and names of Becker and Malinvaud, he said in pathbreaking influence on the development a recent telephone interview .E of econometrics.”5 Published in the journal Economefrica in 1944, the thesis laid out what has become 1S1’s Basis of Forecasting the prize the scientific foundation for testing economic theories and building accurate models. b Passell concluded his article with a ques- Those more accurate models have, in turn, tion: “Have a better idea of who will win? made economic forecasts more accurate. Perhaps you could make a little money on A few days after the announcement of it. David Romer’s straw poll is actually a Haavelmo’s prize, I read a brief article in pool. If a long shot like Trygve Haavelmo the New Yorki’lmeson speculations that had were to win next year, a strategically placed taken place before the new Nobelist was dollar bet might return 40 or 50. “T named. TNoting that few economists would Well, to answer PasSell’squestion, I think have predicted Haavelmo’s selection and 1S1does have as good an idea, if not a bet- that many did not even recognize his name, ter one, of who will win the Nobel Prize in the writer, Peter PasSell,offered up an array economic sciences in the coming years. It of opinions on who might win this year’s is based on the citation fdes of 1S1’sSocial prize in economics, Sciences Citation Index@ (SSCP ). We re- “Here is the morning line, ” he wrote, cently drew up a list of 50 authors who were “unscientifically culled from a dozen phone most cited in the SSCI during the years chats with economists who, sensibly, prefer 1966-1986. to remain nameless. ” The names he listed, In order to identi& highly cited econo- in order, were Gary S. Becker of the Uni- mists among the myriad social scientists versity of Chicago, Illinois; Robert E. whose works are cited in the SSCI, we em- Lucas, also of the University of Chicago; ployed the following mettmlology. First, we Joseph E. Stiglitz of Startford University, isolated all papers cited 50 or more times, California; Ronald H. Cease, emeritus at 1966-1986, that were published in 27 the University of Chicago; Richard A. “core” economics journals. These core Musgrave, emeritus at Harvard; and French journals had been listed in a previous economist Edmond Malinvaud, Collage de essay.g France, Paris. Second, we made a list of all the primary He also mentioned “safe choices, ” a authors of these papers and tabulated cita- group he characterized as “prolific econo- tions for each. The decision to limit the list mists at the top of their specialties. ” His to first authors was a temporary expedi- ‘‘nonexhaustive” list included Jagdish N. ent—the database we used, in fact, was or- Bhagwati of Columbia University, New ganized by primary author only, and ex- York; Peter A. Diamond of the Massachu- panding it to include all authors would have setts Institute of Technology, Cambridge; required considerably more time and ex- Martin Feldstein of Harvard; Robert E. Hall pense. In future essays we hope to use of the Hoover Institution at Stanford; Dale all-author data to identifi the most-cited W. Jorgenson of Harvard; and Edwin S. SSCIeconomists, especially since we’ve ex- Phelps of Columbia. tended the SSCI back through 1956.10 The Passell also spoke to David Romer, an list of names was then ranked by total economist at the University of California, citations. Berkeley, who has been polling economists Third, using this ranked list of names, we for the last seven years on who will win the consulted another 1S1fde that shows, again

84 by primary authors’ names, all works–in- Table 1: Me&+ited economkts io the SSCP, cluding books as well as papers-that were 1966-1986, ranked by tntat primary rurtfror citations. A =mnk. B =rrame. C =birtb y--d@b yWU. cited 10 or more times during this same pe- D =total primary author citations, An asterisk (*) in- riod, 19661986, and totaled the citations for dicates a Nobel laureate. A dagger (f) indicates dcrxased each author. Finally, the names of some three dozen A B c D economists (including Nobel laureates) who ], * /jrr~wKJ (1921) 7,807 had not been previously identified in step 2. * Samuelsun P A (1915) 6,S67 two, but who were suggested by reviewers 3, * sfi~” H A (1916) 5,894 to be highly cited, were checked and their 4. * Friednrarr M (1912) 5,219 5, Becker G S (1930) 4,947 citations counted. Once again, the list was 6, Fama E F (1939) 4,592 ordered by total citations. 7. Feldstein M (1939) 4,512 8. Theil H (1924) 4,207 The result of this work, which entailed a 9. * Stigler G J (1911) 4,150 herculean tabulation effort, appears in Ta- 10, Baumol W J (1922) 4,053 ble 1. 11. * Buchanan J M (1919) 3,428 12. Galbrsith J K (1908) 3,370 I shotdd mention that, after our work was 13, * Tobin J (1918) 3,214 completed, we became aware of a recently 14. tKeynes J M (1883-1946) 3,022 published list of 150 economists ranked by 15. * Modigliani F (1918) 2,898 16. Barro R J (1944) 2,826 citations for the period 1971-1985.11 The 17. t Robinson J (1903-1983) 2,718 origin of these data is the SSCf and the cita- 18. *tHicks J R (1504-1989) 2,650 19. Lucas R E (1937) 2,615 tion data are primary author coortts. The au- 20. Serr A K (1933) 2,5S4 thor, Marshall H. Medoff, California State 21. *tMyrdal G (1898-1987) 2,477 University, Long Beach, restricted himself 22. * Solow R M (1924) 2,286 23, Griliches Z (1931) 2,260 to tabulating citation counts for economists 24, Sargent T J (1943) 2,119 at US universities who were under 65 as of 25, Bowles S (1939) 2,035 1985 and who had not by then won the 26, tHotelling H (1s95-1973) 2,015 27, Mlncr.r J (1922) 2,004 Nobel Prize. We made no such restrictions 28. Cuase RH (1910) 1,950 in our list. I encourage those interested to 29. Nerlove M (1933) 1,942 30. * Debreu G (1921) 1,931 compare these two lists for their similarities 31. Jorgenson D W (1933) 1,929 and differences. 32. Zellner A (1927) 1,830 33. * Schultz T W (1902) 1,816 34. Phelps E S (1933) 1,815 35. Black F (1938) 1,714 Nobel Class Economkts 36. Stiglitz J E (1942) 1,695 37. Olsnn M (1932) 1,662 38. ● Klein L R (1920) I,wl It is important to emphasize again that our 39. Maiinvaud E (1923) 1,625 list is based on pnrnq author data-not 40. ?Lintner J (1916-19s4) 1,623 41, Granger C W J (1934) 1,604 all-author data, Citations to a paper or book 42. Jensen M C (1939) i ,602 with two or more authors were credited to 43. Musgrave R A (1910) 1,564 the first author only. So, if an author has 44. Bbagwati J N (1934) 1,561 45, Alchiarr A A (1914) 1,544 a habit of listing his or her name in other 46. Mansfield E (1930) 1,503 than ftrst position, for whatever reason, that 47. *tKuznets S (1901-1985) 1,502 person would not be identified in this 48. Chow G C (1929) 1,4s3 49, Hirsbieifer J (1925) 1,417 analysis. 50. Chenery H B (1918) 1,3g2 Also, if an economist happens to be a secondary author on one or more highly cited a-in tie bottom half of Table 1. Romer works, that person would not, in this anal- suggested that, because of our method, ysis, receive citation credit. That, some have ‘‘you’ve probably collected fewer than half suggested to us, is the reason why Stiglitz of his citations.”8

85 Using primary author data is, of course, Plainly, some of the names-such as those not the most exhaustive methodology. But of Becker and Lucas-are on everyone’s list an “all-author” study would have had its of potential prizewimers, Both names ap- own limitations-namely, of excluding ci- pear in the top 20. What I find remarkable tations to books, since these are not covered is that every name in Passell’s article, with in the SSCI Source Index. In order to know the exception of two and those he calls’ ‘dark what names are listed as secondary authors horses, ” appears in this automatically on a work, 1S1would have had to have in- generated list. The two that do not, Diamond dexed the work. To exclude citations to and Hall, are ranked 53d and 56th, res~- books would have lessened the influence of tively. Citation data do agree with subjec- an important vehicle for the communication tive opinion to some extent, and they have of ideas in economics, a vehicle much more been shown to correlate with subjective important in the social sciences than in the judgments of influence, sigtilcance, and life or physical sciences. prestige. But to find nearly all the names Another cautionary note: Table 1 is, we suggested by “inside experts” on the 1S1list believe, a good listing of most-cited econo- of the 50 most-cited economists of mists for the period 1966-1986, but we fully 1966-1986 is indeed remarkable. realize that it is not the ultimate list. Because Romer told an interesting story that I have our process of identifying highly cited econ- his permission to repeat. He said that he was omists was not exhaustive-we did not look talking to a colleague about our exercise, up the name of every economist in the and the colleague said, “I think they [the world—names might have been missed that Nobel committee] should just give it to the do belong in the list. The relative rankings guy with the most citations. ” Romer said of those that do appear are accurate, he started to put up some objections about however. how citations are imperfect, when hk col- I should say that ex post ~acto checking league cut him off with ‘‘.. yes, but this of about three dozen or so names resulted would have a much better track record than in relatively few changes to the top 25. One the Nobel committee has had.’ ‘g was John Maynard Keynes, for example, who died in 1946. Keynes is the great eco- Is Haavehno an Exception to the nomic figure of the twentieth century. The Citation Rule? influence of his work is so pervasive that citation of his ideas and specific publications Careful readers will have noticed in Table is deemed by many to be umecessary, a 1 that Haavelmo’s name is not on the list. phenomenon known as obliteration by His total citations, talc W the same way, incorporation. 12 number 237. Does Haavehrto’s selection The appearance of Keynes in this list also disprove the point advanced here—that ci- gives me the opportunity to emphasize that tations are accurate indicators of Nobel his appearance as the 14th most-cited econ- stature? I don’t think so. The Nobel Prize omist of this period most certainly does not in economic sciences was initiated ordy mean that he was the 14th best economist some 21 years ago, in 1%9. There is a large of the paid. We do not attribute increments “backlog” of pioneering economists who of quality to absolute numbers of citations. made their seminal contributions during the His appearance so high in this list, as well 1940sand 1950s, and some even before that, as that of others in the top 20 or 25, merely who are still alive and therefore eligible for indicates that all have had a great impact, recognition by the Nobel committee. as is reflected by the frequent citation of Some have said the committee is in a race their works. against time to recognize this older genera-

86 tion of Nobel class economists. When the cover how a list generated using that meth- backlog is cleared, by prize-giving or by od would compare with Table 1. With these death, it is expected that the annual prize in two we would be in a good position to draw economics will go to more contemporary up a list of most-cited papers and books in figures. Haavelmo is of that older genera- economics. In the meanwhile Clive Crook, tion. His ideas had already become well in- the economics correspondent for the Ecorr- corporated into the fabric of economics dur- omist of London, tells us he is planning a ing the period we examined (1966-1986); multipart series that uses these data as its therefore, explicit citations to his works departure point. IS I am looking forward to were relatively few. his articles on these highly cited, Nobel class Rather, the individuals listed in Table l— economists. if they have not died and have not already won the prize-are potential wimers in fu- ***** ture years. The probability of winning the Nobel Prize for these economists is, if the My thanks to C.J. Fiscus and David past is any indication, very good. Pendlebu~ for their help in the preparation As I noted earlier, we hope to perform an of this essay. all-author citation analysis of most-cited economists. It will then he interesting to dis- oi’m1s1

REFERENCES

1. Pendfebory D. The 1989 Nobel Prixe in medicine: 20 who deserve it. Zbe .$rientisf 3(19):14; 16; 19, 2 Gctober 1989. 2. Sher I H & Wtrffeld E. New touls for improving and evaluating the effectiveness of research. (Yovits M C, Gilford D M, Wilcox R H, Staveley E & Lemer H D, eds. ) Research program eflertiveness.New York: Gordon & Breach, 1966. p. 13546. [Reprinted in: Essays of an inforrndon scientisf. Phifadelphla: 1S1Press, 19S4. Vol. 6. p. 503-13.] 3. GarffeId E. Citation indexing for studying science. Nature 227669-71, 1970. [Reprinted in: L5id, 1977. Vol. 1. p. 133-8.] 4. Cole S, Rubbt L & Cole J R. P&r review md the SUPPO~of acien~. Sci. Amer. 237(4):34-41, Gctober 1977. 5 Royal Swdah Academy of Scferrms. ??risyear k laureate in economics showed how economic theories can be tesfed. 11 October 19g9. 4 p. (Press release.) 6. Haavebno T. The probab~ity approach in econometrics. (Whole issue.) Econornetrica 12(Supp.), 1944. 118 p, 7 Passcff P. The morning Iine on the next Nobel. New York i%ees 18 October 1989. p. D2. 8 Rumer D. Personal communication. 13 Febmary 1990. 9. Diiund A M. The core joumafa of eccmnmics. Current Contents (I ):4-11, 2 January 1989. 10 Garfield E. The new 1956-1%5 Sociid Sciences Citation Me-r, Part 1. Analysis of 19gg research tlonts snd the Cir@”onClassics that made them possible. CurrentContents(41):3-8, 9 Gctoher 1989. 11 Medoff M H. The ranking of eamonrista. 1 &on. Educ. 20405-15, 1989. 12 Merton R K. Soeia[ tAeory d social structure, New York Frw Press, 1%8. p. 27-9; 35-8. 13 Crook C. PerannaJ communication. 5 February 1990.